Simple dataclasses configuration management for Python with hocon/json/yaml/properties/env-vars/dict/cli support.
Requires at least Python 3.9.
# pypi
pip install dataconf
poetry add dataconf
# remote master
pip install --upgrade git+https://github.com/zifeo/dataconf.git
poetry add git+https://github.com/zifeo/dataconf.git
# local repo/dev
poetry install
pre-commit install
import os
from dataclasses import dataclass, field
from typing import List, Dict, Text, Union, Tuple
from dateutil.relativedelta import relativedelta
from datetime import datetime, timedelta
import dataconf
conf = """
str_name = test
str_name = ${?HOME}
dash-to-underscore = true
float_num = 2.2
iso_datetime = "2000-01-01T20:00:00"
iso_duration = "P123DT4H5M6S"
variable_length_tuple_data = [
1
2
3
]
tuple_data = [
a
P1D
]
# this is a comment
list_data = [
a
b
]
nested {
a = test
b : 1
}
nested_list = [
{
a = test1
b : 2.5
}
]
duration = 2s
union = 1
people {
name = Thailand
}
zone {
area_code = 42
}
"""
class AbstractBaseClass:
pass
@dataclass
class Person(AbstractBaseClass):
name: Text
@dataclass
class Zone(AbstractBaseClass):
area_code: int
@dataclass
class Nested:
a: Text
b: float
@dataclass
class Config:
str_name: Text
dash_to_underscore: bool
float_num: float
iso_datetime: datetime
iso_duration: timedelta
variable_length_tuple_data: Tuple[int, ...]
tuple_data: Tuple[Text, timedelta]
list_data: List[Text]
nested: Nested
nested_list: List[Nested]
duration: relativedelta
union: Union[Text, int]
people: AbstractBaseClass
zone: AbstractBaseClass
default: Text = 'hello'
default_factory: Dict[Text, Text] = field(default_factory=dict)
print(dataconf.string(conf, Config))
# Config(
# str_name='/users/root/',
# dash_to_underscore=True,
# float_num=2.2,
# iso_datetime=datetime.datetime(2000, 1, 1, 20, 0),
# iso_duration=datetime.timedelta(days=123, seconds=14706),
# variable_length_tuple_data=(1,2,3),
# tuple_data=('a', datetime.timedelta(days=1)),
# list_data=['a', 'b'],
# nested=Nested(a='test', b=1),
# nested_list=[Nested(a='test1', b=2.5)],
# duration=relativedelta(seconds=+2),
# union=1,
# people=Person(name='Thailand'),
# zone=Zone(area_code=42),
# default='hello',
# default_factory={}
# )
@dataclass
class Example:
hello: str
world: str
foo: List[str]
os.environ['DC_WORLD'] = 'monde'
print(
dataconf
.multi
.url('https://raw.githubusercontent.com/zifeo/dataconf/main/confs/simple.hocon')
.env('DC')
.on(Example)
)
# Example(hello='bonjour',world='monde')
import dataconf
conf = dataconf.string('{ name: Test }', Config)
conf = dataconf.string('name:\n\tvalue: Test', Config, loader=dataconf.YAML) # dataconf.HOCON by default
conf = dataconf.env('PREFIX_', Config)
conf = dataconf.dict({'name': 'Test'}, Config)
conf = dataconf.url('https://raw.githubusercontent.com/zifeo/dataconf/master/confs/test.hocon', Config) # hocon, json, yaml, properties
conf = dataconf.file('confs/test.hocon', Config) # hocon, json, yaml, properties
conf = dataconf.cli(sys.argv, Config)
# Aggregation
conf = dataconf.multi.string(...).env(...).url(...).file(...).dict(...).cli(...).on(Config)
# Same api as Python json/yaml packages (e.g. `load`, `loads`, `dump`, `dumps`)
conf = dataconf.load('confs/test.hocon', Config) # hocon, json, yaml, properties
conf = dataconf.load('confs/test.yaml', Config, loader=dataconf.YAML) # dataconf.HOCON by default
dataconf.dump('confs/test.hocon', conf, out='hocon')
dataconf.dump('confs/test.json', conf, out='json')
dataconf.dump('confs/test.yaml', conf, out='yaml')
dataconf.dump('confs/test.properties', conf, out='properties')
For full HOCON capabilities see here.
PREFIX_VAR=a
PREFIX_VAR_NAME=b
PREFIX_TEST__NAME=c
PREFIX_LS_0=d
PREFIX_LS_1=e
PREFIX_LSLS_0_0=f
PREFIX_LSOB_0__NAME=g
PREFIX_NESTED_="{ name: Test }"
PREFIX_SUB_="{ value: ${PREFIX_VAR} }"
is equivalent to
{
var = a
var_name = b
test {
name = c
}
ls = [
d
e
]
lsls = [
[
f
]
]
lsob = [
{
name = g
}
]
nested {
# parse nested config by suffixing env var with `_`
name: Test
}
sub {
# will have value "a" at parsing, useful for aliases
value = ${PREFIX_VAR}
}
}
Note that when using .env
source, the strict mode is disabled and value might
be casted.
Same as env vars parse (dashes are converted to underscore, e.g. TEST_A
→
--test-a
).
Can be used for validation or converting between supported file formats (-o
).
dataconf -c confs/test.hocon -m tests.configs -d TestConf -o hocon
# dataconf.exceptions.TypeConfigException: expected type <class 'datetime.timedelta'> at .duration, got <class 'int'>